confluence

Confluence MCP provides an interface to interact with Atlassian Confluence through the Model Context Protocol. It exposes a set of tools that let you search for Confluence pages and spaces, retrieve page details and content, create new pages and spaces, update existing pages, manage permissions and metadata, fetch page comments, and list available spaces. You can connect to this MCP server using standard MCP clients and choose between stdio (default) or HTTP-based transport depending on your integration needs. The server is designed to work with typical Confluence Cloud or Server instances, requiring authentication via environment variables to access the Confluence API.

To use it, you typically run the MCP server (for example via Docker as shown in the installation guide) and point your MCP client at the server. In your MCP configuration, you specify the server name and how to start the process. Common tools like Claude Desktop or Cursor can communicate with the server using the stdio transport by providing the appropriate command and environment (for example, pointing to a local confluence-mcp binary or using a Docker container). If you want HTTP-based access, you can enable the HTTP transport by running the server with an HTTP port, and then call the provided HTTP endpoint for MCP operations.

Prerequisites:

• Docker installed on your machine (if using the Docker option)
• Access to a Confluence instance with an API token or suitable credentials
• Optional: Go installed if you build locally; not required for the provided Docker image

Installation options:

Option 1: Docker (recommended for quick start)

• Pull and run the image locally: docker pull ghcr.io/nguyenvanduocit/confluence-mcp:latest docker run --rm -i
  -e ATLASSIAN_HOST=your_confluence_host
  -e ATLASSIAN_EMAIL=your_email
  -e ATLASSIAN_TOKEN=your_token
  ghcr.io/nguyenvanduocit/confluence-mcp:latest

Option 2: Build from source (Go)

• Ensure Go is installed (1.20+ recommended)
• Clone the repository and navigate to it
• Build the binary: go build -o confluence-mcp ./...
• Run locally (example): ./confluence-mcp

Option 3: Download pre-built binaries

• Visit the GitHub Releases page for confluence-mcp
• Download the appropriate binary for your OS
• Make it executable (Linux/macOS): chmod +x confluence-mcp
• Run the binary or place it in your PATH

Environment setup (required):

• ATLASSIAN_HOST: your_confluence_host (e.g., your-domain.atlassian.net)
• ATLASSIAN_EMAIL: your_email
• ATLASSIAN_TOKEN: your_confluence_api_token

Tips and common issues:

• Ensure your Atlassian API token has the necessary permissions for the actions you plan to perform (read pages, create/update content, etc.).
• When using Docker, make sure the host environment variables are correctly passed to the container to authorize with Confluence.
• For HTTP transport, you can run the server with --http_port to expose an HTTP API at http://localhost:8080/mcp. Use a reverse proxy if exposing publicly and consider authentication/authorization.
• If you encounter CQL or content retrieval issues, check that the provided host and token have access to the relevant spaces/pages and that the token format is correct.
• The available tools include search_page, get_page, create_page, update_page, get_comments, and list_spaces. Use the appropriate tool name in your MCP client to perform the operation you want.